WEB DOCUMENT CLASSIFICATION BY CONSIDERINGWORD-CONCEPT AND CONCEPT-ONTOLOGY SUPPORT

摘要

With the rapid growth of Semantic Web, retrievalorientedrntask is among the best-developed in semanticrnweb applications. One of the most important techniquesrnused for this task is document classification. Integratingrnthe background knowledge with the classification algorithmrnto improve the classification efficiency and accuracyrnis a very interesting topic. The paper proposes arnnovel classification algorithm based on Naive Bayesrnapproach. Given a set of ontologies,our task is to classifyrndocuments with the ontologies by calculating the supportrnof the documents to every ontology. Firstly we will obtainrnthe support Pco as the semantically-based supportrnfor a concept to its affiliated ontology through enrichingrnthe concept relevant words in ontology. Secondly, wernwill calculate the support Pwc for each word in the webrndocument, which reflects the support of the word in thernweb document to every ontology concept. Further, wernwill integrate the two kinds of support Pco and Pwc torncalculate the support of the web document to each ontology.rnSo far, we can classify the document to the ontologyrnhaving largest support value or to multi ontologiesrnhaving support values larger than a threshold. The experimentalrnresults show our algorithm is effective forrnontology based classification.